tornado-boilerplate is an attempt to set up an convention for Tornado app layouts, to assist in writing utilities to deploy such applications. A bit of convention can go a long way.
This app layout is the one assumed by buedafab.
Tested with Tornado v3.2
buedafab django-boilerplate python-webapp-etc comrade
The folks at Mozilla working on the next version of AMO were the primary inspiration for this layout.
tornado-boilerplate/
handlers/
foo.py
base.py
lib/
logconfig/
media/
css/
vendor/
js/
vendor/
images/
requirements/
common.txt
dev.txt
production.txt
templates/
vendor/
environment.py
fabfile.py
app.py
settings.py
All of your Tornado RequestHandlers go in this directory.
Everything in this directory is added to the PYTHONPATH when the
environment.py file is imported.
Python packages and modules that aren't really Tornado request handlers. These are just regular Python classes and methods.
Everything in this directory is added to the PYTHONPATH when the
environment.py file is imported.
An extended version of the log_settings module from Mozilla's zamboni.
This package includes an initialize_logging method meant to be called from the
project's settings.py that sets Python's logging system. The default for
server deployments is to log to syslog, and the default for solo development is
simply to log to the console.
All of your loggers should be children of your app's root logger (defined in
settings.py). This works well at the top of every file that needs logging:
import logging
logger = logging.getLogger('five.' + __name__)
A subfolder each for CSS, Javascript and images. Third-party files (e.g. the
960.gs CSS or jQuery) go in a vendor/ subfolder to keep your own code
separate.
pip requirements files, optionally one for each app environment. The
common.txt is installed in every case.
Our Fabfile (see below) installs the project's dependencies from these files.
It's an attempt to standardize the location for dependencies like Rails'
Gemfile. We also specifically avoid listing the dependencies in the README of
the project, since a list there isn't checked programmatically or ever actually
installed, so it tends to quickly become out of date.
Project-wide templates (i.e. those not belonging to any specific app in the
handlers/ folder). The boilerplate includes a base.html template that defines
these blocks:
title - Text for the browser title bar. You can set a default here and
append/prepend to it in sub-templates using {{ super }}.
site_css - Primary CSS files for the site. By default, includes
media/css/reset.css and media/css/base.css.
css - Optional page-specific CSS - empty by default. Use this block if a page
needs an extra CSS file or two, but doesn't want to wipe out the files already
linked via the site_css block.
extra_head - Any extra content for between the <head> tags.
header -Top of the body, inside a div with the ID header.
content - After the header, inside a div with the ID content.
footer - After content, inside a div with the ID footer.
site_js - After all body content, includes site-wide Javascript files. By
default, includes media/js/application.js and jQuery. In deployed
environments, links to a copy of jQuery on Google's CDN. If running in solo
development mode, links to a local copy of jQuery from the media/ directory -
because the best way to fight snakes on a plane is with jQuery on a plane.
js - Just like the css block, use the js block for page-specific
Javascript files when you don't want to wipe out the site-wide defaults in
site_js.
This needs to be tested with Tornado's templating language. A quick look at the documentation indicates that this basic template is compatible, but none of our Tornado applications are using templates at the moment, so it hasn't been tested.
Python package dependencies loaded as git submodules. pip's support for git repositories is somewhat unreliable, and if the specific package is your own code it can be a bit easier to debug if it's all in one place (and not off in a virtualenv).
At Bueda we collect general webapp helpers and views in the separate package
comrade and share it among all of our applications. It is included here as an
example of a Python package as a git submodule (comrade itself should't be
considered part of this boilerplate - while it might be useful, it's much less
generic).
Any directory in vendor/ is added to the PYTHONPATH by environment.py. The
packages are not installed with pip, however, so if they require any
compilation (e.g. C/C++ extensions) this method will not work.
Modifies the PYTHONPATH to allow importing from the apps/, lib/ and
vendor/ directories. This module is imported at the top of settings.py to
make sure it runs for both local development (using Django's built-in server)
and in production (run through mod-wsgi, gunicorn, etc.).
We use Fabric to deploy to remote servers in development, staging and production environments. The boilerplate Fabfile is quite thin, as most of the commands are imported from buedafab, a collection of our Fabric utilities.
The main Tornado application, and also a runnable file that starts the Tornado server.
A place to collect application settings ala Django. There's undoubtedly a better way to do this, considering all of the flak Django is taking lately for this global configuration. For now, it works.
If you have improvements or bug fixes:
- Fork the repository on GitHub
- File an issue for the bug fix/feature request in GitHub
- Create a topic branch
- Push your modifications to that branch
- Send a pull request
- Bueda Inc.
- Christopher Peplin, peplin@bueda.com, @peplin